A Neural Network Based Defect Prediction Approach for Virtual High Pressure Die Casting

Abstract

Prediction of defects is important to effective process planning for high pressure die casting (HPDC). Current computer aided engineering (CAE) methods of defect prediction are widely used by experienced engineer in industry. However, it is hard for novices to image and understand the underlying relationship between the process and defects. To bridge the gap between training and onsite applications, this paper present a neural network based defect prediction approach (DPA) for virtual HPDC, and details of the DPA development and its implementation in VR are explained. Moreover, a Virtual HPDC Lab is developed as the case study to demonstrate the functionality of DPA proposed, and the result survey verified that the virtual lab with DPA is very much effective for learning and training of HPDC.